rikhil bhavnani saumitra jhasaumitra/papers/gandhisgiftepsjfinal.pdfas the violence of the 1942 quit...

THE ECONOMICS OF PEACE AND SECURITY JOURNAL BHAVNANI and JHA, Gandhi’s Gift p. 76Vol. 9, No. 1 (2014) | doi:10.15355/epsj.9.1.76

The Economics of Peace and Security Journal — ISSN 1749-852X — http://www.epsjournal.org.uk © EPS Publishing, 2014. All rights reserved. For permissions, email: [email protected]

GANDHI’S GIFT: LESSONS FOR PEACEFUL REFORM FROM INDIA’S STRUGGLE FOR

DEMOCRACY

RIKHIL BHAVNANI and SAUMITRA JHA

Rikhil Bhavnani is Assistant Professor of Political Science at the University of Wisconsin, Madison, WI, USA. He may bereached at [email protected]. Saumitra Jha is Associate Professor of Political Economy at the Graduate School of Business,Stanford University, Stanford, CA, USA. He may be reached at [email protected].

AbstractWe examine the potential and limitations of nonviolent civil disobedience through the lens of the evolution of an iconicsuccess: India’s struggle for democratic self-rule. We summarize evidence consistent with a theoretical framework thathighlights two key challenges faced by nonviolent movements in ethnically diverse countries. The first challenge, that offorging a mass movement, was met through the brokering of a deal that took advantage of an external shock (in this case, theGreat Depression) to align the incentives of disparate ethnic and social groups toward mass mobilization in favor of democracyand land reform. The second challenge, that of keeping the mass movement peaceful, was accomplished throughorganizational innovations introduced by Mohandas Gandhi in his reforms of the constitution of the Congress movement in1919-23. These innovations took the movement from one dominated by a rich elite to one organized on the principle ofself-sacrifice. This permitted the selection of future leaders who could then be trusted to maintain nonviolent discipline inpursuit of the extension of broad rights and public policy objectives.

We have a power, a power that cannot be found inMolotov cocktails, but we do have a power.

Power that cannot be found in bullets and in guns,but we do have a power.

It is a power as old as the insights of Jesus ofNazareth and as modern as the techniques of MohandasK. Gandhi.

—Dr. Martin Luther King1

There were moments in the twentieth century whenactivists believed that a new technology of politicalorganization—that of mass nonviolent civil

disobedience—yielded a new power for affecting institutionalchange around the world. Drawing upon age-old religioustraditions common to many of the major faiths of the world, theidea of eschewing violent action in favor of nonviolentresistance was not new. Yet, modern techniques of civildisobedience incorporated new organizational ideas that havebeen credited with a number of remarkable successes. Theseinclude the ceding of democratic rights to 30 million SouthAsians by the British Empire in the 1930s and the civil rightsmovement in the United States in the late 1950s and the 1960s.

However, nonviolent civil resistance can also fail. Modernscholars of civil resistance point to the issue of maintainingnonviolent discipline in the face of provocation as an important

missing piece in our understanding of how to make civilresistance work.2 And on the ground, as historic episodes suchas the violence of the 1942 Quit India movement, the race riotsthat followed the 1950s and 1960s U.S. civil rights movement,and the 2011 Arab spring and the battles in Cairo’s TahrirSquare demonstrate, movements that begin peacefully can beprone to rapid breakdown into violence that further facilitatesrepression.

In this overview article, we summarize research in progressthat examines the potential and limitations of nonviolent civildisobedience through the lens of the evolution of an iconicsuccess: India’s struggle for democratic self-rule.3 We begin bysketching a theoretical framework to delineate conditions underwhich nonviolent mass movements can succeed and when theymay fail. We show that even in environments conducive tononviolent effectiveness, such as when policymakers facegreater costs to violent crackdowns of nonviolent movementsrelative to their nonviolent counterparts, there remain two keychallenges faced by nonviolent movements. The first is thechallenge of mass mobilization: Nonviolent movements, moreso than violent movements, are only effective when they arelarge in scale. This problem becomes even more challengingwhen attempting to mobilize across ethnic groups, which oftendiffer in their policy preferences and suffer from weakenedabilities to communicate, coordinate, and sanction acrossethnic lines.



The second challenge lies in overcoming the enhancedtemptations faced by members of large mobilized groups toturn violent, whether to secure short-term gains from mobaction or in response to manipulation by agents who stand togain from political violence. We provide general patterns fromcross-campaign data which are consistent with thesechallenges.

Motivated by these patterns, we discuss how despite thefact that both these challenges were particularly accentuated inthe South Asian case, they were overcome. We argue that thefirst challenge, that of forging a mass movement, wasaccomplished through the brokering of a deal that tookadvantage of external shocks (in this case, the GreatDepression) to align the incentives of disparate ethnic andsocial groups toward mass mobilization in favor of democracyand land reform. The second challenge, that of keeping themass movement peaceful, was met through organizationalinnovations introduced by Mohandas Gandhi in his reforms ofthe constitution of the Congress movement in 1919-23. Thesetook the movement from one dominated by a rich elite to oneorganized on the principle of self-sacrifice, selecting futureleaders who could then be trusted to maintain nonviolentdiscipline in pursuit of the extension of broad rights and publicpolicy objectives. We characterize these reforms in light of theeconomics of club goods, describing how Gandhi’s reformscreated incentives that ironically mimic those used by terroristorganizations and extreme religious cults to developwithin-group trust.4

We then outline new and emerging evidence documentingthe historical importance of organizational innovations andincentives in shaping the nature and success of the Indianindependence movement, drawing upon a range of hithertolargely untapped sources, including recently releasedintelligence reports gathered by the British, as well as internalCongress correspondence. We conclude by arguing that a key,but hitherto mostly neglected, aspect of Gandhi’s Gift—theexample of nonviolence applied to India’s independencestruggle—lies in understanding these organizationalinnovations. We discuss how these findings relate tocontemporary movements for democracy.

Theoretical frameworkIt is useful to sketch a theoretical framework to help understandthe potential and limitations of nonviolent civil disobedienceas a technology, and the role played by mobilization andleadership. Suppose that individuals in a society differ in their“type,” defined as the cost to them of engaging in violentaction. Factors such as military training and combat experiencemay lower such costs, or conversely there may be cultures of

honor that make it more costly for some to “turn the othercheek” when confronted with violence. Also assume thatindividuals can choose at some cost to join a movement, andsubsequently, whether within a movement or not, to engage inviolence.5

Further, assume that a movement becomes more intense ifmore people are mobilized and if they possess commonlyaccepted organizational leadership. The benefits to movementintensity from having a leader are relatively small when fewpeople are people mobilized, but increase as the numbers rise.Thus, leaders complement followers in forging a more intensemovement, be it through directing individuals toward commonobjectives, organizing logistics, or other relevant activity.

Suppose there are two possible types of potential benefitfrom increasing the intensity of movements: private benefitsthat only accrue to movement members, and public benefitsthat accrue to all. Further suppose that leaders of a movement,if they exist, can decide whether to permit or penalize violentacts within the movement. A more intense movement thatpermits violence can be used both to influence public policyand public goods provision, but also allows local private gainsto both its leaders and followers through the ability to loot theproperty of other individuals or groups more effectively. Incontrast, define a nonviolent movement—one that prohibitsviolence—as different from other movements only in that itsstrategy either requires its leader to expel members who engagein violence or imposes sufficient penalties on violent membersto check their behavior. Both types of movements may allowsome private gains to members, such as status benefits withinthe organization. However, all other things being equal,movements that permit violence have greater potential forprivate gains than those that constraint violence.

Finally, suppose that after these private gains are realized,an external decisionmaker seeks to minimize the cost imposedby the movement either by granting a policy concession, byignoring the campaign, or by violent repression. In ourmotivating case, the decisionmaker encompasses the Britishgovernment, of course. Suppose that the decisionmaker getsthe same fixed payoff from a policy concession, such as

We examine the potential for and limitations of nonviolentcivil disobedience through the lens of the evolution of an iconicsuccess: India’s struggle for democratic self-rule. We identifytwo key challenges: First, that of forming a nonviolent massmovement in the first place and, second, how to keep itnonviolent once formed. We find that both unexpectedexternal events as well as organizational innovation withinGandhi’s mass movement were important. We employ uniquedatasets and draw tentative lessons for cases other than India.



granting the democratic franchise. In contrast, the costs of bothignoring a campaign and of engaging in violent repression arenot fixed but are increasing in a movement’s intensity, withdoing nothing being the lower cost option for dealing withsmall movements, and violent repression becoming cheaper forthe decisionmaker when confronted with larger and moreintensive movements.

A crucial condition for the relative effectiveness of violentand nonviolent campaigns comes from changes in the relativecost to a decisionmaker from choosing to violently repress acampaign as that campaign’s intensity rises. We will say thatthe environment is in the audience state when the relative costof violently repressing a nonviolent campaign becomes higherthan violently repressing violent campaigns as campaignintensity rises, and in the isolated state when repressingnonviolent campaigns violently becomes relatively cheaperthan repressing violent campaigns as campaign intensity rises.

Audience and isolated states can be thought of ascorresponding to the role played by the international systemand the media. In isolated environments where no one outsidethe movement and the State are aware of violent repression, itis arguably cheaper to violently suppress groups committed tononviolence than those who are willing to retaliate. In contrast,in the audience state, greater reputational costs or otherpenalties imposed on the decisionmaker may exist for violentlysuppressing a nonviolent campaign as opposed to one that hascommitted violence.

An audience state may be more likely when a pro-democracy international hegemon exists or where individualshave access to freer media, as has been increasingly the casewith today’s social networking technologies. In our motivatingcase of Indian independence, the British, particularly during thecivil disobedience movement and the second world war, wereable to censor much of the Indian media yet were deeplyconcerned about American opinion as well as that of some oftheir own constituents.6

This setup makes explicit a number of the assumptions webelieve necessary for a nonviolent campaign to work andsuggests some of the challenges that such campaigns may face.Working backwards, note first that, in this setup, the onlyadvantage that nonviolent campaigns have over campaigns thattolerate violence arises in the higher costs decisionmakers facein violently repressing them rather than granting policyconcessions. This advantage holds only if the campaign issufficiently intensive (i.e., sufficiently large and organized) sothat, had it been violent, the decisionmaker would prefer toviolently repress the campaign over ignoring it or granting thepolicy concession and if the decisionmaker faces audiencecosts that make suppressing nonviolent campaigns costlier than

suppressing violent campaigns.Thus, nonviolent movements require both sufficient scale

and the presence of audience costs to repression to be moreeffective at influencing public policy than movements thatpermit violence. In addition, in our framework, nonviolentcampaigns also face greater difficulties in both recruitment andorganization relative to similar campaigns that do not penalizeviolence. First, nonviolent movements lack some of the privategains that violence can permit. From the perspective of privategains, a movement that does not penalize violence will be morefavorable to its leaders and potentially more attractive to abroader range of types of followers than an otherwise similarnonviolent movement, as a nonviolent movement limits theprivate gains from looting or expression that can be hadthrough violent action. In essence, it is easier to mobilize amob that contains opportunists who use a movement to settlescores and to loot than to mobilize a group that is willing togive up the prospect of short-term gain and willing to “turn theother cheek” despite provocation.7

A second aspect to note is that keeping movementspeaceful becomes harder as they grow. As movements increasein size and intensity, both leaders and followers have anincreasing incentive to allow violence for private gains, evenif this may compromise the chances of subsequently gaining apublic policy concession. Thus, leaders of movements seekingto impose nonviolent discipline must not only forego their owntemptation to permit violence; increasingly they must organizeeffective penalties on their followers to prevent a nonviolentmovement from unraveling into violence.

This is particularly hard as nonviolent leaders actually havefew options for penalizing violence. While leaders that permitviolence can also potentially use the threat of violence toimpose discipline, the worst a nonviolent leader can do to anyspecific individual is to make that member indifferent betweenparticipating or not, i.e., to expel him or her. But because thepolicy gains are public goods, members of large movementsmay have an incentive to turn to violence, knowing that evenif they are expelled from the movement, they will benefit fromany policy concessions regardless. All that individual membersstand to lose are the private benefits from staying within thenonviolent movement.

But another penalty available is for a leader to refuse to(continue to) direct the movement. Given the complementaritybetween leaders and followers, this will reduce the movement’sintensity and may reduce the probability of public policy gainsfor all. If there are few commonly acceptable leaders and thecomplementarity between leaders and followers is sufficientlylarge, this otherwise weak threat may shift a marginally violenttype of follower from preferring violence to maintaining



nonviolent discipline. Indeed, this approachwas one exercised on a number ofoccasions by Mahatma Gandhi.

With such weak penalties available forthose who might choose to engage inviolence, personnel selection becomescritical for nonviolent movements. A keychallenge is to identify and mobilizeenough of a select group of followers whoare willing to forego the temptation ofviolence in favor of national objectives,and an even more select group of leaders toforego even greater such temptations, sothat the movement is sufficiently intensiveand yet able to maintain nonviolentdiscipline.

Ethnic differences can make itparticularly difficult to support theseincentives. Ethnic groups may havedifferent policy preferences, reducing thebenefits from a single policy concessionand thus to a cross-ethnic movement as awhole. This makes it difficult for amovement to grow in scale beyond anethnic group. Differing policy preferencescan also accentuate the challenge offollowers accepting a common leader. The potential lack of areplacement, stemming from the threat of resignation we justdiscussed, may however strengthen the power of any leadersthat do emerge.8

To summarize: Nonviolent movements can be moresuccessful than violent movements at achieving policyoutcomes. But this depends upon their ability to mobilize enmasse and on whether external decisionmakers face audiencecosts. Moreover, leaders and followers in movements thatbecome large face heightened private temptations to exploit amovement through its increased capacity for violence. Thusnonviolent movements must depend more heavily than othermovements on the selection of leaders, as well as followers,who are willing to forego the temptations that an increase inthe size of a movement brings. Finally, the problems of findingsuch recruits, and of developing a commonly acceptableleadership to discipline them, are accentuated in sociallydivided societies.

General patternsUsing cross-campaign data, it is a useful first check to seewhether our theoretical framework matches general patterns ofsuccess and failure of violent and nonviolent campaigns. Thus,

Figure 1 employs the NAVCO 2.0 dataset of 250 nonviolentand violent campaigns between 1945 and 2006. It comparescampaigns within the same country, showing how differentsizes of violent and nonviolent campaigns in a year affect theprobability that a campaign successfully achieves its policyobjectives. Campaign sizes are measured on the horizontal axesof the three-panel figure.

The panel on the left differentiates between violent (darkervertical bars) and nonviolent (lighter vertical bars) movements.The probability that a campaign achieves 100 percent of itsstated objectives is, if anything, lower for nonviolentcampaigns than for violent campaigns with less than 100,000or so members. But an increase in scale brings a rise in theirrelative effectiveness.

When one separates campaign-years (within the samecountry) between those that attracted high international mediacoverage (middle panel) from those with less coverage (rightpanel), nonviolent movements are no more successful thanviolent ones under little or moderate media coverage butstrikingly more likely to be successful when they exceed500,000 or so members and the international media take aninterest.

Consistent with our theory, then, the relative policy

Figure 1: Effectiveness of violent and nonviolent campaigns, by size and internationalmedia coverage, 1945-2006. Notes: Estimated coefficients of size on the probability ofsuccess, controlling for country fixed effects. Standard errors clustered at the country level.N=1594 campaign-years, spanning 250 campaigns with >1000 members that held goals ofregime change, independence, or self-determination. “Success” defined as campaignachieving 100% of stated goals. Size is a lower bound estimate. Left: All-campaign years;Middle: Campaign-years that attracted high international coverage; Right: Campaign-yearsthat attracted no, low, or moderate international coverage. Source: Authors’ calculationsbased on NAVCO 2.0 dataset (Chenoweth and Lewis, 2013).



effectiveness of nonviolent campaigns appears to rise withmovement size and the audience costs imposed viainternational media. Yet, as our theory also outlines, this maybe only part of the overall picture: Although increased scalemay make a nonviolent campaign more effective, if itmaintains discipline, it may also lead to an increasedtemptation for members of nonviolent campaigns to turnviolent. Table 1 compares campaigns from the samecountry-year in their relative probability of turning violent andshows that large nonviolent campaigns, and campaigns thatstart off with attempts to embrace groups across ethnic lines,are more likely to turn violent.

The example of South Asian independenceThe broad patterns discussed thus far make the achievementsof South Asian independence movements even moreremarkable. The 1947 independence of what would becometoday’s India, Pakistan, and Bangladesh was the first majorreversal of a process of colonization by Europeans that hadbeen continuing since the liberation of the states of LatinAmerica in the aftermath of the Napoleonic wars (Figure 2).This made South Asia a prominent example for civil rights andindependence movements yet to come around the world.

How and why then did a broad coalition of South Asiansfrom across ethnolinguistic and economic lines push fordemocratic self-determination? How was this coalition largelysuccessful at maintaining a nonviolent mass movement andwhy did it also, at times, fail? In Bhavnani and Jha (2014a) weprovide the first systematic evidence on the relative importanceof economic, cultural, and organizational factors in mobilizingthe Indian subcontinent’s remarkably diverse population intoone of the world’s first nonviolent mass movements in favor ofdemocratic self-government. We exploit a range of hithertolargely untapped subnational (administrative, district-level)data, assembling novel data on mobilization, including terroristacts between 1893 and 1936, votes and turnout in the firstprovincial elections in 1936, newly declassified intelligencereports on violent insurrection and nonviolent protest duringthe civil disobedience movement in 1930 and the GreatRebellion of 1942 against British rule, and internal Congressparty accounts of membership and mobilization. Our workdraws upon original archival correspondence and records fromthe Indian Office Records and National Archives in the UnitedKingdom, the National Archives of India in New Delhi, andthe papers of the All-India Congress Committee and BombayPradesh Congress, both now housed in the Nehru MemorialLibrary.

While this work is still in progress, the emerging evidencepoints to an intriguing set of patterns that motivate a major re-

evaluation of the lessons to be learned from India’s freedomstruggle. As discussed, our theoretical framework points to twokey challenges of nonviolent mobilization in the South Asiancontext: mobilizing sufficient numbers of followers acrossethnic and societal lines, and selecting leaders and followers

Table 1: Size, cross-ethnic campaigns, and transitions fromnonviolence to violence

(1) (2) (3)Outcome: Campaign turns Nonviolent to violent = 1

violent = 1 No change = 0 Violent to nonviolent = –1

Probit OLS OrdereddF/dX Probit

Campaign size 0.293* 0.038* 0.454*>100,000 (0.145) (0.017) (0.226)

Campaign spans 0.003* 0.001 0.002ethnic lines (0.001) (0.001) (0.003)

Country FE Yes Yes YesYear FE Yes Yes YesSample Country-years Full Full

with variationR-squared 0.15Observations 217 1,594 1,589

Notes: Column (1) includes only campaigns in country-years inwhich some campaigns transition from primarily nonviolent toviolent and other campaigns in the same country-year do not.Columns (2) and (3) includes all campaign-years. Column (3)re-parameterizes the outcome to rank campaigns that transition tononviolent practices (-1), retain existing practices (0), or turnviolent (+1). Standard errors clustered at the country level. *significant at 10%; ** 5%; *** 1%. Source: Authors’ calculationsbased on NAVCO 2.0 dataset (Chenoweth and Lewis, 2013).

Figure 2: World trends in decolonization. Note: The vertical linemarks the independence of India and Pakistan in 1947. Source:Authors’ calculations based on Cross-National Time-Series DataArchive.



that can impose and accept nonviolentdiscipline. In our working paper andbroader book project (2014a; 2014b), weargue, and provide evidence, that economicshocks from the Great Depression(1929-1933) were important forencouraging mobilization across ethniclines, while Gandhi’s reforms of theCongress movement (1919-23) wereparticularly important for the selection ofleaders that could be trusted tosubsequently impose nonviolent discipline.

A common view of the history ofIndia’s independence struggle is shaped byits ultimate success, that of a sequence ofsuccessful mobilizations, masterminded byGandhi, that built upon themselves andculminated with independence in 1947.Table 2 summarizes our contrastingreinterpretation of the struggle andsuggests the generalizable lessons itprovides. First, we observe that nonviolentmass mobilization in India was only trulysuccessful in achieving substantial policyconcessions in one campaign: the civildisobedience movement (1930-1932) thatbegan with a tax revolt in the form ofGandhi’s Salt March, and was called offonly after the Viceroy, Lord Irwin, agreedto a pact that allowed salt concessions,released civil disobedience prisoners,recognized the Congress party as a keyinterlocutor for the political fate of India,and paved the way for the Government of India Act of 1935.This Act granted the first broad democratic franchise in SouthAsian history, expanding the number of people with the rightto vote from 0.65 percent in 1934 to 14 percent of the adultpopulation, or 30.1 million people, in elections that began to beimplemented a year later. The Act also provided substantialautonomy over local public goods to new provinciallegislatures. This victory arguably also made subsequentconcessions, including independence, become much morelikely. We argue that the civil disobedience movementbenefitted from having solved both of the key challenges ofnonviolence: first, incentives for mass mobilization werecreated by economic shocks that swayed agriculturalists—akey constituency—in Congress’ favor, and second, Gandhi’sorganizational reforms, which helped impose nonviolentdiscipline.

In contrast, attempts by Indian nationalists to influenceBritish policy prior to Gandhi, and Gandhi’s own other greatmobilization attempts, the non-cooperation movement of1919-22 and the Quit India movement of 1942, failed toachieve any policy concessions. These failures are, however,deeply informative about the conditions under whichnonviolent movements can achieve success.

Prior to 1919, it was unclear how important the IndianNational Congress would be for Indian politics. The Congresswas an elite group, financed and dominated by affluentEnglish-speaking professionals, particularly lawyers andbusinessmen. This group pushed for greater Indian consultationon government within the British Empire. In parallel with thisorganization were regional groups of more extreme nationalistswho, through newspapers and terrorist acts, conducted acampaign for independence. A number of these regional

Table 2: Economic shocks and organizational reform in India’s independencestruggle

Movement Leadership &organization

Incentivesfor massmobilization

Achieves aims

Nationalismbefore Gandhi,1885-1918

No: Leadersselected by elitestatus

No: Ethnicmovements;elite movements

No: No policyconcessions;violentrepression(Sedition Act1919)

Mobilizationattempt 1:Khilafat/non-cooperationmovement,1919-1922

Yes: Gandhi’sReforms:Leaders selectedby costlysacrifice fornonviolent action

No: Aimsmisaligned withagriculturalsector andacross ethnicmovements

No: No policyconcessions;Hindu-Muslimviolence

Mobilizationattempt 2: Civildisobediencemovement,1930-1932

Yes: Nonviolentleaders selected

Yes: GreatDepressionallows coalitionformation withagriculturalsector

Yes: Gandhi-Irwin Pact 1932;Gov’t India Act1935 provideslocal autonomyand extendsfranchise

Mobilizationattempt 3: QuitIndia movement:August 1942

No:Comprehensivedetention ofnonviolentleaders

Yes: Coalitionestablished

No: Campaignturns violent;violentrepression; nofurther policyconcessions

Source: Bhavnani and Jha (2014b).



nationalist groups were following the playbooks of Europeannationalism. For example, the prominent nationalist“Lokmanya” Bal Gangadhar Tilak in Poona used hisnewspaper, Kesari [Saffron], to propagate nationalism, mostnotably by adopting Hindu symbols, such as promoting a minorfestival for the god Ganesha into a major religious and politicalevent, and promoting the exemplar of the Maratha ruler ShivajiBhosle who had fought the Mughal empire. Both of thesesymbols appear to have been aimed at forging a Hindu“imagined community” and coincided with Hindu-Muslimrioting. Kesari’s circulation remained limited regionally. Evenamong Hindus, Tilak’s attempts to propagate Shivaji, a localruler, as a symbol of nationalist resistance in Bengal, on theother side of the subcontinent, met with little success. Bengalitself was viewed by the British as a key center for nationalismin this period, mainly focused on an urbanized affluent elite,known as the bhadralok, or “respectable ones.” In fact,between 1893 and 1918, of the 276 seditious incidentsrecorded in the Sedition Committee report of 1918, 71 percentoccurred within a single province, Bengal, with the provincesof Punjab (12 percent) and Bombay (4 percent) the nearestrivals. More than 90 percent of the events involved attemptedor actual violence, including armed theft, murder, andbombings, while other acts included seditious speeches and thedissemination of articles.9

Gandhi’s reformsThis first phase—combining a small and almost exclusivelyelite-led Congress agitating peacefully even while low-levelviolent sedition, bomb-throwing, and other acts of terrorismwere perpetrated, mainly by those outside the Congressorganization—lasted until around 1919. This was a periodwhen grievances were particularly accentuated, since theBritish reneged upon Viceroy Lord Montagu’s declaration thatIndia would receive dominion status (i.e., self-government) inreturn for military support during the first world war. Instead,in response to the Sedition Committee report, the governmentimposed a series of laws aimed at curtailing “sedition” andlimiting public assembly (Sedition Act, 1919).

It was then that Gandhi returned from South Africa andintroduced the techniques of nonviolent civil disobedience tothe subcontinent as well as pushing for, and obtaining, a broadreform of the Congress organization. The Congress wentthrough a large-scale reorganization during the period 1919-23,adopting a new constitution that changed its steering body, theAll-India Congress Committee, from one that in 1919 wasdominated by elites, particularly lawyers (65 percent), to onethat by 1923 was representatively elected from districtCongress committees that had emerged to span British India

(Table 3).10 The organization extended down to the taluk (anadministrative subdivision) and village level, where eachvillage with more than five Congress members was entitled toa committee and to send delegates to the taluk and districtcommittees. Following Gandhi’s reforms in 1920, members ofthese committees of the Indian National Congress wererequired to give up any positions that they enjoyed with theBritish government, and lawyers in the leadership had to giveup the practice of law in British courts. Engaging in nonviolentcivil disobedience could lead to arrest and prison time, ofteninvolved hard labor, and became an organization-specificinvestment.

Important work, beginning with Laurence Iannaccone, haspointed to costly investment in group-specific identity, such assacrifice, as a form of screening device in cults and clubs,where a key objective is to maintain small group size andscreen for trustworthy members. This logic has been foundparticularly applicable to violent mobilization along ethnic andreligious lines, including among organized crime and terroristorganizations. Violent acts and crimes that reduce a member’soutside options often play the role of a group-specificinvestment. In contrast, we suggest that civil disobedienceprovides a dimension of public sacrifice, includingincarceration and “turning the other cheek” when faced withbrutal law enforcement, that has a similar clubs good structurebut can transcend sectarian and ethnolinguistic differences andalso facilitate nonviolence. In India, political incarcerationprovided a movement-specific investment that was potentiallyopen to all, regardless of their initial cultural and resourceendowments. As Table 3 suggests, the number of memberswhose professions were not identifiable (and thus were more

Table 3: Composition of delegates to the All-India CongressCommittee, 1919-1923

Year 1919 1920 1921 1922 1923Members 160 162 163 338 338

% Lawyers 65.0 64.8 50.9 23.1 21.3% Journalists 7.5 7.4 11.7 6.2 8.6% Businessmen 6.9 7.4 8.6 5.6 4.4% Doctors 4.4 3.7 4.3 4.7 4.7% Landowners 4.4 4.9 2.5 2.4 1.5% Teachers 3.8 3.1 4.3 3.0 3.0% Others 5.0 5.6 6.1 6.8 6.5% Congress workers 0.0 0.0 0.6 4.4 8.0% Unknown 3.1 3.1 11.0 43.8 42.0

Source: Authors’ calculations based on Krishna (1966, p. 424).During Gandhi’s reforms of the Congress movement, theproportion of lawyers fell considerably while those not classified(likely non-elites) rose.



likely to be non-elites) increased from 3.1 percent to 42.0percent over a five-year period.11

The selection of leaders, rather than being based on elitestatus, instead became one based on sacrifice. In the languageof our framework, Gandhi’s reforms created a process forscreening of potential leaders who either are nonviolent types,received sufficient private gains from being part of themovement, or internalized its public policy objectives. Theseleaders, in turn, could later be entrusted with large-scalemobilization despite its temptations.

Having a cadre of leaders is beneficial to nonviolence, but,as discussed, nonviolent leaders’ ability to maintain disciplineis weak. Thus, selection of followers is also important. Thecostly signals that even the lowest-level Congress volunteerhad to provide begin to make more sense in this context. InNovember 1921, the Congress sought to centralize the processof volunteer recruitment, issuing general instructions whichrequired all volunteers to give up any military-style uniforms,banned the carrying of any weapons, ruled out the enlistmentof any “known to be a bad character,” and required volunteersto make a pledge of obedience and of nonviolence. Congressleaders and volunteers were also enjoined to spend at least anhour every day spinning cloth.12

The Khilafat and non-cooperation movements, 1919-1922The reorganization of the Congress coincided with Gandhi’sfirst attempt at mass mobilization in India. This occurred inpart to challenge the Sedition Act that made protests illegal, butalso to create solidarity with the Muslim community, whichfeared for the fate of the Ottoman Sultan, the titular Caliph ofSunni Islam. Pan-Islamic nationalists, many of whom wereconcentrated in India, formed the Khilafat (Caliphate)conference to pressure Britain into maintaining the Caliph’sauthority. Arguing that national unity required massmobilization of both Hindus and Muslims on this issue, Gandhiasked “how can twenty-two crore [ten million] Hindus havepeace and happiness if eight crore of their Muslim brethren aretorn in anguish?” As we discuss in Bhavnani and Jha (2014b),despite having developed a subcontinental organization, thecombined non-cooperation—Khilafat movement was small-scale, particularly compared to those that came after, andalthough Gandhi himself attracted large crowds, the movementitself failed to attract much concrete civil disobedience outsideIndia’s towns.13

An important test of the Congress organization came on 4February 1922. A joint Khilafat/non-cooperation movementnonviolent protest in the town of Chauri Chawra was firedupon by police, leading to the deaths of three protestors. Thedemonstrators became a mob, which burned down the police

station, killing 22 policemen. Gandhi immediately called a haltto the non-cooperation movement and was effective inimplementing this nationwide.

Ultimately, however, the non-cooperation movement andthe Khilafat agitation failed to achieve significant reforms.14

Further, there was a breakdown of cooperation between theMuslim Khilafat movement and the mainly Hindu Congress.Consistent with our framework, local politicians appeared tohave taken advantage of the new era of mobilization, and thisled to the first major wave of civil Hindu-Muslim rioting inSouth Asian history (Figure 3).

While Congress, under Gandhi’s influence, developed thebasic organization and techniques for nonviolence in this firstattempt at nationwide mass civil disobedience, an underlyinglack of aligned incentives, both between the Khilafat andCongress movements, and between the non-cooperationmovement and many that it had hoped to mobilize, may haveplayed a key role both in the limited success of the movementand in the ethnic conflict that resulted where its mobilizationefforts had at first met success.

The civil disobedience movement, 1930-1932What the Congress leadership needed was a set of incentivesthat would make self-rule attractive. One problem was thatIndia’s abundant factor was agriculture, and thus the majorityof Indians employed in the agricultural sector arguablybenefitted from the higher relative prices for agricultural goodsavailable abroad under Britain’s policy of trade openness ascompared to the platform of protectionism promised by theindustrialist-supported Congress.

As Figures 4 and 5 suggest, the Great Depression had alarge negative effect on a key benefit of Empire: the benefitsfrom trade.15 Farmers began to switch from growing non-foodcrops for export to growing food. In this manner they reduced

Figure 3: Satyagraha movements and Hindu-Muslim riots. Source:Jha (2013) and Wilkinson (2005).



Figure 5: Ratio of non-food to food crop acreage. Source:Agricultural censuses, 1929-1935.

their reliance on world demand and thus on their reliance onboth the trade and extension services provided by variousintermediaries, including landlords, and the trade opennesspolicy of the British. The Congress party appears to haveseized upon the reduced enthusiasm of farmers for British ruleand promised them land redistribution from the now redundantlandlord class. Jawaharlal Nehru became the first majorpolitician to speak on the need for land reforms in 1929, a topicthat became a key part of the Congress platform not longafterward. It was soon after the Great Depression struck—on26 January 1930, and thenceforth celebrated as IndependenceDay—that Congress officially changed its platform fromself-government within the British empire to Purna Swaraj andinitiated its next great attempt at mass mobilization, the civildisobedience movement. It is this movement, enjoying both theCongress’ organization and, for the first time, the alignedincentives of the poor in both rural and urban sectors, thatproved the most successful of all of Gandhi’s campaigns.These led, as we discussed, to the granting of provinciallegislatures and to India’s first elections with a substantialfranchise in 1936.16

We can use the election outcome to check whetherCongress mobilization did in fact respond to depression-eraexport shocks. Panels a and b of Figure 6 (on the next page)show the relation of the value of exports between 1923 and1933 to the relative support for the Congress and the MuslimLeague. Note, in Panel b, that the Muslim League, the onlyother national party allowed to contest the elections (theCommunists were not permitted to participate), has very lowsupport even among Muslim constituencies, and does bestamong the winners from the Great Depression, i.e., in areasthat saw only small adverse, or even positive, effects of thedepression on the value of their export goods. The Congress,in contrast, did worst both among the relative economicwinners and the extreme losers but garnered substantial supportamong communities in which the depression’s adverse effectswere more moderate. The voting outcome suggests that thosehit worst by the depression, particularly those who had failedto change away from exportables, were not coordinated intosupport for the opposition. This is consistent with the lack ofattraction that Congress’ autarkic platform might yield to thosewho could not easily substitute away from export goodsproduction. Instead of being a rebellion of those facing thehardest times, support for Congress came from districts thatwere more insulated from the depression or able to adjust moreeasily to domestic production.

The Quit India movement, 1942One might expect that, having built a new mass coalition in

favor of independence—and a cadre of leaders committed tononviolent discipline by 1932—the Congress would now enjoygreat success. Indeed, Gandhi’s third attempt at nonviolentmobilization was proclaimed on 9 August 1942, with themaximalist objective that the British should “Quit India.”

However, the Quit India movement failed. DraconianBritish countermeasures were employed to decapitate theCongress organization. Ironically, these countermeasures nowallow us to empirically evaluate the role of leadership inmaintaining nonviolence. On 3 August 1942, a “most secret”telegram was dispatched to all Provincial Governors. It detailedthat in the event that the Congress should pass the Quit Indiaresolution at its upcoming meeting, the Governors should“arrest ... all individuals whom they consider competent andlikely to attempt to organise and launch mass movement. Noindividual will be arrested merely as a member of unlawfulassociation general object being not to fill the jails but to limitthe number of arrests to those regarded as essential fordislocation of the Congress organisation.” In a synchronizedaction, which commenced within a few hours after the protests

Figure 4: India’s trade with the British Empire. Source: Data fromMitchell (2003).



began, the British arrested and imprisoned an estimated 60,000Congress cadres, including the entire national leadership. Thisprocess was made easier by the fact that many leaders had beengathered in Bombay at the All-India Congress Committeemeeting. Heavy war-time censorship exercised at the time haslong concealed the full magnitude of the mobilization.17

By for the first time bringing together all declassifiedreports on the Quit India movement from each province, webelieve that we can shed new light on its dynamics. Whilemuch work remains to be done, the basic patterns areilluminating. Figure 7 reveals that on the day of the Quit Indiaresolution—8 August 1942—there was a spike in nonviolentprotests across the subcontinent. The overnight arrest of muchof the Congress leadership removed a key element inmaintaining nonviolent discipline. Consistent with our theory,this led to an overnight spike in violent actions; actions thatwould continue to erupt intermittently for the next few months,even as there was a consistent fall in nonviolent mobilization.

In internal war-time Congress correspondence, the role thatthe arrest of leadership played in causing the movement to turnviolent is also evident. The internal Report of August 1942Struggle, penned during war-time by an anonymous memberof the Bombay Pradesh Congress Committee (PCC) explains:

For the sake of proper understanding of the situation it mustbe made clear here that the Congress machinery that cameto assume charge of affairs in the City was not appointedby the Provincial Congress Committee or by any othercompetent Congress authority. The PCC had not drawn outany plan since it had not instructions from the WorkingCommittee. What subsequently happened was moreaccidental than a result of any previous planning. It isevident from what followed that those in charge of theCongress machinery had no special partiality for strictnon-violence as interpreted by Mahatma Gandhi. This willexplain some of the incidents that occurred during thecampaign towards the end of 1942.18

The change in the nature of the Quit India protests fromnonviolent to violent following the arrests of the Congressleadership likely reduced the costs to the British to respondwith violent repression. The suppression of the movementrequired 8 British brigades and 57 Indian battalions. The RoyalAir Force even machine-gunned civilians from the air.19

American and Chinese nationalist pressure to grant concessionsabated, and Congress leaders would spend the rest of the warin jail. They would emerge four years later to find that theirsupport, particularly among Muslims, had eroded and theirposition to speak for all ethnicities diminished. While the

02

04

06

08

0%

Con

gre

ss V

ote

, 1

936

-100 -50 0 50% Export Shock, 1922-1933

02

04

06

08

0%

Mu

slim

Le

agu

e V

ote

, 193

6

-100 -50 0 50% Export Shock, 1922-1933

Figure 6: Export shocks and votes, 1922-1933. Notes: Localpolynomial smooths (solid lines) weighted by district populations,with 95% confidence intervals (dashed lines). Marker sizes denoterelative populations. Square markers are for Muslim-majoritydistricts. Source: Bhavnani and Jha (2014b).

Figure 7: Nonviolent and violent protests, 1941 and 1942. Source:Authors’ calculations based on intelligence reports for each province.



1. Transcribed from King: A Filmed Record ... Montgomery toMemphis. Produced by Ely Landau, 1970, Library ofCongress/Kino Lorber, 00:01:25-00:01:49.

successes of the civil disobedience movement arguably madeIndia’s independence feasible and likely, the failure of the QuitIndia movement may have done the same for the subsequentpartition of the subcontinent into India and Pakistan.

DiscussionThis article summarizes some of our research in progress thatexamines both the potential and the limitations of nonviolentcivil disobedience through the lens of the evolution of an iconicsuccess, India’s struggle for democratic self-rule. Ourtheoretical framework highlights the need for an audience toimpose costs on those who would use violent repression ofnonviolent movements, as well as on two key challenges facedby nonviolent movements in ethnically diverse countries, evenif such audiences do exist. The first is the challenge of massmobilization across ethnic lines. The second lies in overcomingthe enhanced temptations faced by members of large mobilizedgroups to turn violent, whether to secure short-term gains frommob action or in response to manipulation by agents who standto gain from political violence. These challenges appear tomatch general patterns from cross-campaign data.

Sustaining a nonviolent movement requires identifying andmobilizing sufficient numbers of potential followers willing toforego temptations to engage in violence in favor of nationalobjectives as well as leaders who forego even greater suchtemptations so that the movement is sufficiently intense whilestill maintaining nonviolent discipline.

Our theory suggests two design elements for those whoseek peaceful reform. The first design element is to choose thepublic objective that would be most attractive to large numbersof potential nonviolent types rather than attracting those thatmay prefer violence. While sectarian and otherethnically-delimited objectives may be relatively moreeffective in mobilizing people—partly out of the threat thatother ethnic groups may mobilize instead—such ethnicmobilization is also relatively conducive for violence and limitsscale to that ethnic group. In contrast, campaigns against therestrictions on political rights or the imposition of taxes that arecommon to all may encourage fewer potentially violent typesto join and have the potential to mobilize across ethnic lines.The Congress’ use of common economic and political issues,using a platform of the salt tax, protectionism, land reform, anddemocracy to mobilize a nonviolent cadre of Indians in thesuccessful civil disobedience movement in 1930 contrastsfavorably to the ethnic mobilization and lack of aligned interestof the Khilafat movement and the parallel non-cooperationmovement in the 1920s.

A second, oft-overlooked, key to ensuring nonviolentdiscipline lies in the organizational incentives of the

movement. We argue that in the early 1920s, although lackinga broadly attractive public objective, Gandhi’s reformstransformed India’s independence movement. The Congresswent from a group dominated by elites—many with strong tiesto the existing regime and thus good outside options—to a newcadre structure that allowed leadership candidates from bothexisting elites and non-elites to advance within the movementif they were willing to sacrifice potential private objectives fornonviolent public objectives. While the required loss of officeswithin the existing regime reduced the benefits for defection,Gandhian requirements of self-sacrifice—from the mundane,such as yarn spinning, to the physically dangerous, such as thecourting of arrest for civil obedience and engaging in hardlabor in British jails—helped individuals signal theirtrustworthiness to accept nonviolent discipline and thus assumeleadership within the movement. This signal, in contrast tosimilar screening devices by cults and terrorist organizations,was explicitly nonviolent and nonsectarian. Those Congressmembers that made such organization-specific investmentsearly, in the 1920s, when the Congress had little chance ofassuming government, became leaders in the 1930s, permittingthe key success of the civil disobedience movement. Theimportance of this selection of leaders in keeping themovement nonviolent was put into sharp relief in the final greatmobilization of 1942, when the synchronized arrest of 60,000Congress leaders led to a rapid breakdown of nonviolentdiscipline.20

Early Indian nationalists sought to adopt the playbook ofterror and sectarianism employed by nationalists in the West.Yet, a key part of Gandhi’s gift may lie in his reorganization ofa movement that took the screening devices of religious cultsand terror organizations and instead found a method to selectnonviolent leaders and followers. While social movementsoften emphasize the importance of sheer numbers in attainingpolicy objectives, movements that seek peaceful reform maygain from trading off scale with the selection of followers thatare willing to sacrifice short-term private gains for long-termpublic ones. Furthermore, the dimension of visible sacrifice,particularly the use of courting arrest for acts of civildisobedience that Gandhi used are not India-specific but mayprovide avenues for finding trustworthy leadership elsewhere.

South Asia’s struggle for independence has long been anexample for freedom struggles around the world. There mayyet be more that it can teach us.

Notes



2. See, for example, the very useful overviews in Schock(2013, p. 284) and Chenoweth and Cunningham (2013). Oncivil disobedience generally, see Helvey (2004); Sharp (2005);Schock (2005); Shaykhutdinov (2010); and Chenoweth andStephan (2011).

3. Bhavnani and Jha (2014a).

4. Mimic: See Iannaccone (1992); Berman and Laitin (2008).

5. Theoretical framework: See Bhavnani and Jha (2014a;2014b). Factors: See, e.g., Akerlof and Kranton (2000);Grosjean (2011); Jha and Wilkinson (2012).

6. Pro-democracy hegemon: Boix (2011). The British: See forexample the 5 February 1930 confidential dispatch fromGerard Campbell, British Consul-General in San Francisco, toSir Esme Howard, the British Ambassador in Washington earlyin the civil disobedience movement (IOR L/P& J/6/1989): “...I am sincerely convinced that the more friendly relations whichhave recently been established between Great Britain and theUnited States may be endangered if irresponsible Indians areallowed to go around the country telling half-educatedAmericans of the oppression and brutalities suffered by theirfellow countrymen, for these stories will be believed unlesssomeone who can rightfully claim to have the requisiteknowledge is at hand to check or deny them. If the Indianextremists follow up their campaign with a plea for financialassistance, there is no knowing how much they may be able tocollect from the sort of people who seem to derive enjoymentfrom subscribing to any unworthy cause and especially fromthose who, from their youth up, have been taught to be hostile,rather than friendly disposed towards anything British ...” Forsimilar concerns during the Quit India movement, see Sir R.Campbell to Mr. Eden, Telegram L/PO/6/102s: f 7 (Mansergh,1976, p. 659). Many accounts also emphasize the role of thepresence of the world media in mitigating British repression ofthe Dandi Salt March in 1930, and of television in shapingbroader public opinion on the violent repression of civil rightsactivists in the United States.

7. We differ in this respect from existing scholarship (e.g. ,Chenoweth and Stephan, 2011) that suggests that nonviolentmovements are cheaper to join than are violent movements, asless is required of nonviolent agents. Our account insteademphasizes that effective nonviolent movements are both moredifficult to mobilize and to organize, because committedfollowers do not stand to gain from the benefits of violentmobilization and movement leaders are unable to leverage thepayoffs to violence to maintain nonviolent discipline.

8. Different policy preferences: See, e.g., Alesina, Baqir, andEasterly (1999). Another issue is that when there are ethnicgroups with differing policy preferences, individuals maychoose to join different movements with competing policyobjectives. Here, strategic complementarities may exist:Mobilization and violence by one ethnic movement mayincrease the incentives to mobilize by members of otherethnicities to avoid being caught off-guard. In such competitiveenvironments, members of movements that permit violence

may once again see growth in membership that nonviolentcross-ethnic movements may not. For examples of this type ofethnic security dilemma from the Balkans, South Asia, andelsewhere, see Posen (1993); Kaufmann (1996); Fearon (1998);Jha and Wilkinson (2012).

9. Congress was an elite group: Krishna (1966). Imaginedcommunity: See, e.g., Anderson (1983); Jaffrelot (2008). Metwith little success: Sedition Committee (1918, p. 19). In fact:Authors’ calculations based on Sedition Committee (1918).

10. Also see Krishna (1966).

11. Iannaccone (1992). Particularly applicable: See, e.g.,Berman and Laitin (2008).

12. The pledge reads: “I shall faithfully and diligently carry outall instructions received from my superiors. I shall observepledge of nonviolence in word and in deed, and shall inculcatespirit of nonviolence amongst others. I shall regard this pledgeas binding upon me so long as policy of nonviolence iscontinued by Nation. I shall run all risks attendant uponperformance of my duty.” IOR/L/PJ/6/1778, File 7359Decision of the All India Congress as to Organisation andControl of National Volunteers.

13. Gandhi quote: Lelyveld (2011, p.157).

14. Kemal Ataturk would depose the Ottoman Sultan, relievingthe British of the responsibility and removing the main reasonfor the Muslim mobilization.

15. In Figure 4, note that following the repatriation of warmateriel between 1918-1921, India returned to similar levels ofimports and exports to the prewar period. Also note thatrelative to these benchmarks, trade with the British Empiretook a sharp dip as the Great Depression struck in 1929-1930.

16. Nehru: See Malaviya (1954).

17. Quote: Government of India, Home Department toSecretary of State, Telegram, L/ P & J/8/597:118-19.

18. Report of August 1942 Struggle. Bombay PradeshCongress Committee papers. Nehru Memorial Library SpecialCollections, 1942/BPCC/File 29.

19. As Linlithgow wrote to the Secretary of State for India: “Ifyou have any trouble in the debate [in Parliament] aboutshooting from air, it may be worthwhile mentioning that inmany cases this action was taken against mobs engaged intearing up lines on vital strategic railways in areas whichground-floor forces could not reach ... But this is not true of allcases in which firing occurred from aircraft ...” –Viceroy toSec. State (4 October 1942).

20. We have assumed that certain societies are likely to be inthe audience state largely for exogenous reasons. However, asdiscussed, the presence of an audience is critical for nonviolentmovements, and they may be able to influence such attentionon some margin. In our book project (Bhavnani and Jha,2014b), we discuss strategies famously employed by Gandhito attract international media coverage, from the use ofsymbols, his own correspondence and publicity outlets, to the



slow pace of the Dandi Salt March to allow media attention togrow.

ReferencesAkerlof, G.A. and R.E. Kranton. 2000. “Economics and

Identity.” Quarterly Journal of Economics. Vol. 115, No.3, pp. 715-753.http://dx.doi.org/10.1162/003355300554881

Alesina, A., R. Baqir, and W. Easterly. 1999. “Public Goodsand Ethnic Divisions.” Quarterly Journal of Economics.Vol. 114, No. 4, pp. 1243-1284.http://dx.doi.org/10.1162/003355399556269

Anderson, B. 1983. Imagined Communities: Reflections on theOrigin and Spread of Nationalism. London: Verso.

Berman, E. and D. Laitin. 2008. “Religion, Terrorism andPublic Goods: Testing the Club Model.” Journal of PublicEconomics. Vol. 92, pp. 1942-1967.http://dx.doi.org/10.1016/j.jpubeco.2008.03.007

Bhavnani, R. and S. Jha. 2014a. “Forging a Non-Violent MassMovement: Economic Shocks and OrganizationalInnovations in India’s Struggle for Democracy.”Mimeo,Stanford University.

Bhavnani, R. and S. Jha. 2014b. “Gandhi’s Gift: On thePromise and Limitations of Non-Violent CivilDisobedience.” Book project in progress.

Boix, C. 2011. “Democracy, Development and InternationalSystem.” American Political Science Review. Vol. 105, No.4, pp. 809-828.http://dx.doi.org/10.1017/S0003055411000402

Chenoweth, E. and M. Stephan. 2011. Why Civil ResistanceWorks: The Strategic Logic of Nonviolent Conflict, NewYork: Columbia University Press.

Chenoweth, E. and K.G. Cunningham. 2013. “UnderstandingNonviolent Resistance: An Introduction.” Journal of PeaceResearch. Vol. 50, No. 3, pp. 271-276.http://dx.doi.org/10.1177/0022343313480381

Chenoweth, E. and O.A. Lewis. 2013. “Unpacking NonviolentCampaigns: Introducing the NAVCO 2.0 Dataset.” Journalof Peace Research. Vol. 50, No. 3, pp. 415-423.http://dx.doi.org/10.1177/0022343312471551

Fearon, J.D. 1998. “Commitment Problems and the Spread ofEthnic Conflict,” in D.A. Lake and D.S. Rothschild, eds.,The International Spread of Ethnic Conflict: Fear,Diffusion, and Escalation. Princeton, NJ: PrincetonUniversity Press.

Grosjean, P. 2011. “A History of Violence: The Culture ofHonor as a Determinant of Homicide in the U.S. South.”http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1917113 [accessed 12 March 2014].

Helvey, R. 2004. On Strategic Nonviolent Conflict: ThinkingAbout the Fundamentals. Boston, MA: Albert EinsteinInstitute.

Iannaccone, L.R. 1992. “Sacrifice and Stigma: ReducingFree-riding in Cults, Communes and Other Collectives.”Journal of Political Economy. Vol. 100, No. 2, pp. 271-291.

http://dx.doi.org/10.1086/261818Jaffrelot, C. 2008. “The Politics of Processions and

Hindu-Muslim Riots,” pp. 280-308 in S.I. Wilkinson, ed.Religious Politics and Communal Violence. New Delhi:Oxford University Press.

Jha, S. 2013. “Trade, Institutions and Ethnic Tolerance:Evidence from South Asia.” American Political ScienceReview. Vol. 107, No. 4, pp. 806-832.http://dx.doi.org/10.1017/S0003055413000464

Jha, S. and S. Wilkinson. 2012. “Does Combat ExperienceFoster Organizational Skill? Evidence from EthnicCleansing during the Partition of South Asia.” AmericanPolitical Science Review. Vol. 106, No. 4, pp. 883-907.http://dx.doi.org/10.1017/S000305541200041X

Kaufmann, C. 1996. “Possible and Impossible Solutions toEthnic Conflict.” International Security. Vol. 20, No. 4, pp.136-175.http://dx.doi.org/10.2307/2051999

Krishna, G. 1966. “The Development of the Indian NationalCongress as a Mass Organization, 1918-23.” Journal ofAsian Studies. Vol. 25, No. 3, pp. 413-430.http://dx.doi.org/10.2307/2051999

Lelyveld, J. 2011. Great Soul: Mahatma Gandhi and HisStruggle with India. New York: Alfred Knopf.

Malaviya, H.D. 1954. “Land Reforms in India.” New Delhi:Economic and Political Research Department, All IndiaCongress Committee.

Mansergh, N. 1976. “The Transfer of Power 1942-47.”London: Her Majesty’s Stationery Office.

Mitchell, B.R. 2003. International Historical Statistics: Africa,Asia and Oceania, 1750-2000. Basingstoke, England:Palgrave Macmillan.

Posen, B.R. 1993. “The Security Dilemma and EthnicConflict.” Survival. Vol. 35, No. 1, pp. 27-47.http://dx.doi.org/10.1080/00396339308442672

Schock, K. 2005. Unarmed Insurrections: People PowerMovements in Nondemocracies. Minneapolis: University ofMinnesota Press.

Schock, K. 2013. “The Practice and Study of CivilResistance.” Journal of Peace Research. Vol. 50, No. 3,pp. 277-290.http://dx.doi.org/10.1177/0022343313476530

Sedition Committee. 1918. Report. Calcutta: Government ofIndia Press. Justice S.A.T. Rowlatt Presiding.

Sharp, G. 2005. Waging Nonviolent Struggle: 20th CenturyPractice and 21st Century Potential. Boston, MA:Extending Horizons Books.

Shaykhutdinov, R. 2010. “Give Peace a Chance: NonviolentProtest and the Creation of Territorial AutonomyArrangements.” Journal of Peace Research. Vol. 47, No.2, pp. 179-191.http://dx.doi.org/10.1177/0022343309353109

Wilkinson, S.I. 2005. “Wilkinson Dataset on Hindu-MuslimViolence in India, 1600-1950.”

rikhil bhavnani saumitra jhasaumitra/papers/gandhisgiftepsjfinal.pdfas the violence of the 1942 quit...

Documents